Anthropic launches Claude Haiku 5.5 at 90% lower prices

Anthropic launched Claude Haiku 5.5 on October 7, 2026, introducing its cheapest, fastest, and most capable small model to date. The model targets high-volume and speed-sensitive workloads such as document summaries, database queries, live customer support chats, and autonomous web browsing.

For prompts up to 100,000 tokens, developers pay $0.10 per million input tokens and $0.50 per million output tokens. That matches the rate OpenAI introduced on September 22 for GPT-6 Luna. Compared with Haiku 4.5 rates of $1 and $5, base pricing is 90% lower, while prompts above 100,000 tokens receive a 50% discount. Because roughly 90% of requests fell below that prompt threshold, Anthropic estimates average user savings at near 75%.

The release also marks the first Haiku model with an adjustable effort setting to balance cost and reasoning depth. In addition, Anthropic halved the cache-read price for Claude Sonnet 5.5 to $0.10 per million tokens. The rollout follows Opus 5.5 on September 22 and Sonnet 5.5 on September 28, completing the three Claude 5.5 releases. Opus 5.5 was the company's first launch after CEO Dario Amodei published an essay urging the industry to moderate rapid capability advances.

On the OSWorld 2.1 benchmark, Haiku 5.5 posted a 72.4% success rate compared to 48.9% for Luna. On Terminal-Bench 4.0, Haiku 5.5 completed 39.2% of tasks on the first attempt, against 16.4% for Luna, 0% for Haiku 4.5, and 70.6% for Sonnet 5.5. On the GDPval-AA v2.1 professional Elo scale covering 44 occupations, Haiku 5.5 achieved a score of 1620, beating Luna at 1437 and Haiku 4.5 at 735.

Haiku 5.5 is available on the Claude platform, Amazon Web Services, Google Cloud, and Microsoft Azure as claude-haiku-5-5. Anthropic is also providing monthly API credits of $100 for Max 5x subscribers, $200 for Max 20x, and up to $500 pooled for Team plans.